INTRODUCTION
On 24 March 2020 , lockdown was imposed on the whole country including DELHI.This exploratory data analysis aims to see how lockdown affected Air quality of DELHI by comparing the data of 2020 with the previous year of the same dates. First we try to see broadly how AQI values were affected . Then we see how each pollutant was affected . And finally try to statistically test our observation.
AIR QUALITY DATA FROM 2019
AIR QUALITY DATA FROM 2020
- The data set have 17 columns where the column names represent the pollutants.
- X is the id of our data , city is delhi and date of recording the data.
- The AQI bucket has the labels associated with AQI value of that particular date
COMPARISON OF AQI BUCKET DATA
- Here we observe that we have more number of days where AQI has remained moderate and satisfactory.
- Air quality was more in the breathable range in 2020 compared to 2019
COMPARISON OF AQI VALUES
- The boxplot also confirms that AQI has been lower in 2020 as compared to 2019.
- The data has become skewed in 2020 ( more towards lower side)
- According to IQR range we have three outliers in our 2020 data.
POLLUTANT SPECIFIC ANALYSIS
- We observe that since about the middle of march the concentration of the pollutants have deacreased. The deacrease is visible with most of the pollutants.
- Ozone and Xylene haven’t followed this pattern.
BOXPLOT OF POLLUTANTS
- We observe a clear reduction in concentration in most of the pollutants.
- The observations of our line plot are supported from the boxplots.And Ozone and Xylene haven’t seen such loss.
- We see that variance of concentration of pollutants have increased in the year 2020 and data seems to have become skewed towards lower side.
PROBABLITY DENSITY PLOTS
- Most of the data doesn’t seem to follow normal distribution. Which is confirmed by Shapiro-Wilk test for normality.
- And as shown by the boxplot the data of 2020 has skewed towards left side.
- Only NO2 , SO2 and O3 data from 2019 seems to follow normal distribution
- To make distribution unskewed we try log transformation
PROBABLITY DENSITY PLOT (DIFFRENCE)
- The diffrences density plot looks unskewed .
- We will use shapiro wilk test to test normality of our data to see whether we can use t-test or not.
Shapiro–Wilk test for Normality
| Transformation |
PM2.5 |
PM10 |
NO |
NO2 |
NOx |
NH3 |
CO |
SO2 |
O3 |
Benzene |
Toluene |
Xylene |
| Unchanged |
0.4540317 |
0.2270518 |
0.1000007 |
0.4800837 |
0.5705924 |
0.02091078 |
0.8016024 |
0.9460928 |
0.0146819 |
0.9026673 |
0.3333945 |
1.79429e-11 |
| log Transformed |
0.6147983 |
0.1478432 |
0.324549 |
0.004784697 |
0.4289944 |
0.06082354 |
0.9647551 |
0.3627109 |
0.06802442 |
0.9640188 |
0.1021412 |
NaN |
| sqrt transformed |
0.3269246 |
0.2778903 |
0.711266 |
0.06815003 |
0.786053 |
0.04417748 |
0.9533469 |
0.5789124 |
0.08860538 |
0.9831196 |
0.4682499 |
0.003832592 |
We applied it on the unchanged ,log transformed and sqrt transformed data results are as shown above
- Test on unchanged data shows that except NH3, O3 and Xylene failed the test.
- On log transformed data most of the data passed the test except NH3,Xylene
- On sqrt transormed data only Xylene and NH3 failed the test
- Therefore we will log sqrt transform the Ozone data before applying t test on it
- And for NH3 and Xylene we will do will WILCOX TEST
t test on our data
| PM2.5 |
PM10 |
NO |
NO2 |
NOx |
CO |
SO2 |
O3 |
Benzene |
Toluene |
| 1.959126e-09 |
5.530422e-12 |
1.211396e-07 |
2.196962e-18 |
3.941476e-11 |
5.878381e-15 |
9.734808e-13 |
0.3574597 |
1.764066e-14 |
1.695036e-18 |
- From above results we can reject our null hypothesis for the above pollutants excepts Ozone.
- Barring Ozone , we have enough statistical evidence to state that covid 19 induced lockdown reduced the concentration of the above listed pollutants.
- For our non normal pollutant data (NH3 , Xylene) we will do wilcox test
WILCOX TEST FOR NON NORMAL DATA
| PM2.5 |
PM10 |
NO |
NO2 |
NOx |
NH3 |
CO |
SO2 |
O3 |
Benzene |
Toluene |
Xylene |
| 3.22092e-10 |
1.466224e-10 |
4.267188e-11 |
2.561156e-15 |
8.065608e-13 |
0.0008484531 |
6.187262e-14 |
3.462018e-11 |
0.2500416 |
4.976796e-14 |
2.236049e-16 |
8.510146e-06 |
- Based on the results above we can reject our null hypothesis for all the pollutants except Ozone.
- Only in case of Ozone we have not enough statistical evidence to say that covid 19 lockdown reduced its concentration.
CONCLUSION
- AQI deacreased drastically in since lockdown was imposed
- Most of the pollutant concentration came down except ozone and Xylene.
- Our hypothesis test reveals that the average concentration of all pollutants was lower compared to previous year except for Ozone.